FinOps in Practice: Cutting AWS Costs Without Slowing Down Engineering

In the rapidly evolving cloud landscape, organizations are constantly seeking ways to optimize their cloud spending without compromising on performance and scalability. This case study delves into how adopting FinOps practices led to a significant reduction in AWS costs by 35%, highlighting the importance of cost visibility, rightsizing, choice between Reserved Instances and Savings Plans, leveraging spot instances, optimizing auto-scaling, and fostering a cost-aware culture.

The journey to cloud cost optimization is often riddled with complexities, primarily due to the dynamic nature of cloud services pricing and the continuous deployment of resources. A mid-sized tech company, facing spiraling AWS costs, embarked on a FinOps implementation journey to gain better control over their cloud spend. This case study outlines the strategies employed and the tangible results achieved.

The first step in the FinOps journey was to establish a clear baseline of current cloud spending. This was achieved by utilizing AWS Cost Explorer and AWS Budgets for detailed insights into where the dollars were flowing.

{
  "view": "timeSeries",
  "stacked": false,
  "metrics": [
      ["AWS/EC2", "BlendedCost"]
  ],
  "timePeriod": {
      "start": "2023-01-01",
      "end": "2023-12-31"
  }
}

This JSON snippet for AWS Cost Explorer API retrieves the blended cost of EC2 instances over the year, providing a clear view of spending patterns.

The analysis revealed significant spending on underutilized resources and instances running 24/7 without necessity. This insight led to the first actionable strategy: rightsizing.

Rightsizing involves adjusting the size of an instance or service to match the workload. The company utilized AWS Compute Optimizer to identify recommendations for downsizing or changing instance types to more cost-effective alternatives.

aws compute-optimizer get-recommendation-summaries --account-ids "<AccountID>"

By implementing these recommendations, the company achieved an immediate 10% reduction in compute costs.

Reserved Instances (RIs) were initially considered due to their substantial discounts over on-demand rates. The commitment to a specific instance type for 1 or 3 years in exchange for lower costs seemed appealing.

However, analysis showed that AWS Savings Plans offered greater flexibility by applying discounts to any EC2 usage regardless of instance family, size, region, OS, or tenancy, provided that the commitment was in terms of dollars per hour over 1 or 3 years.

After a detailed cost-benefit analysis, the company opted for a combination of Compute Savings Plans for predictable compute costs and EC2 Instance Savings Plans for specific instances with predictable usage, achieving a further 15% cost reduction.

Spot Instances, which allow users to take advantage of unused EC2 capacity at significant discounts, were identified as an ideal solution for workloads with flexible start and end times. The company implemented a strategy to use Spot Instances for stateless and elastic workloads such as batch processing jobs.

import boto3

client = boto3.client('ec2')

response = client.request_spot_instances(
    SpotPrice='0.03',
    InstanceCount=1,
    Type='one-time',
    LaunchSpecification={
        'ImageId': 'ami-0abcdef1234567890',
        'InstanceType': 'm5.large',
    }
)

This Python script demonstrates how to request Spot Instances, resulting in a 20% decrease in the cost of these workloads.

The company also optimized its auto-scaling groups to ensure that scaling actions were closely aligned with actual demand. By fine-tuning the scaling policies and integrating AWS Auto Scaling with AWS CloudWatch metrics, the company ensured that resources were scaled down during low-usage periods and scaled up only when necessary, leading to further cost savings.

{
    "AutoScalingGroupName": "MyAutoScalingGroup",
    "PolicyName": "TargetTrackingScaling",
    "PolicyType": "TargetTrackingScaling",
    "TargetTrackingConfiguration": {
        "PredefinedMetricSpecification": {
            "PredefinedMetricType": "ASGAverageCPUUtilization"
        },
        "TargetValue": 50.0
    }
}

This JSON configuration for AWS Auto Scaling ensures that the instance count in the auto-scaling group adjusts based on average CPU utilization, maintaining a balance between performance and cost.

The implementation of FinOps practices wasn't solely a technical endeavor but also involved cultural changes within the organization. The company launched training sessions and workshops to raise awareness about cloud cost management and encouraged teams to adopt a cost-aware mindset.

To foster accountability, the company utilized AWS cost allocation tags to assign cloud costs to specific projects or departments. This transparency ensured that every team was aware of their spending and motivated to optimize their resource usage.

{
    "ResourceId": "instance-id",
    "Tags": [
        {
            "Key": "Project",
            "Value": "ProjectA"
        }
    ]
}

This JSON snippet illustrates how to assign a cost allocation tag to an instance, enabling detailed tracking of expenses per project.

The company established a dedicated FinOps team responsible for continuously monitoring, analyzing, and optimizing cloud costs. This team leveraged AWS Trusted Advisor and other third-party tools to identify cost-saving opportunities regularly.

By implementing the strategies outlined above, the company achieved a substantial 35% reduction in AWS costs within a year, translating into significant savings. More importantly, the shift towards a cost-aware culture ensured that these savings were sustainable over the long term.

The case study of this mid-sized tech company demonstrates the profound impact that implementing FinOps practices can have on cloud cost optimization. By gaining visibility into cloud spending, rightsizing resources, making informed decisions between Reserved Instances and Savings Plans, leveraging Spot Instances, optimizing auto-scaling, and fostering a cost-aware culture, organizations can achieve substantial cost savings while maintaining, or even enhancing, their cloud infrastructure's efficiency and scalability. The journey towards cloud cost optimization is ongoing, but with the right strategies and a dedicated approach, significant improvements are within reach.